workflows/flowcraft: restore current schema compatibility - #6
Conversation
The v0.2 Flowcraft resources retained fields from an older runtime shape that the current GizClaw declarative Workflow decoder rejects. - Remove unsupported execution, persistence, and inline tool configuration - Restore direct RuntimeProfile model aliases and valid conversation starters - Preserve later token limits and Journey voice alias improvements Generated with [Codex](https://github.com/openai)
1 similar comment
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There was a problem hiding this comment.
🤖 OpenAI PR review
Reviewed commit: 4accc291ea
Model: gpt-5.6-terra
Reasoning effort: medium
Duration: 47s
Input tokens: 184,974
Cached input tokens: 147,858
Cache write tokens: 37,092
Cache hit ratio: 79.9%
Output tokens: 2,212
Reasoning tokens: 1,188
Total tokens: 187,186
One lifecycle regression found.
1 inline finding published.
idy
left a comment
idy
left a comment
There was a problem hiding this comment.
Review result: no blocking findings
Reviewed exact head 4accc291eabcb7415fe50641493bb62e75e13459 against issue #5 and GizClaw origin/main (20f3282e, including #528).
Confirmed:
- all 1,031 YAML files parse and all 10 Workflow resources strictly decode with the current generated
apitypes; - the three repaired Flowcraft resources match the current built-in schema-compatible forms, while preserving the intended v0.2 token and voice-alias changes;
- a candidate archive built from this head closes successfully against Desktop RuntimeProfile/default, including chatroom, pet-care, the three selected Flowcraft workflows, and their model/voice/resource bindings;
git diff --checkpasses and all live PR checks are green.
I also rechecked the resolved case_solved thread. The removed node depended on schema-invalid inline tools/tool_names, so the published v0.2 resource could not have provided a working notification. Routing directly to write_case_audit matches the schema-compatible v0.1 resource and the current GizClaw built-in workflow. A real external lifecycle event would require the separately defined ToolResource/executor/authorization contract excluded by #5, so I do not consider that a regression in this compatibility fix.
No additional bug found. This PR is ready from my review perspective; publishing immutable v0.2.1 remains the post-merge release step.
🤖 OpenAI PR review completed with an errorReason: The workflow stopped unexpectedly before reporting a more specific reason. |
There was a problem hiding this comment.
🤖 OpenAI PR review
Reviewed commit: b30fe39ab8
Reviewed range: a622026ca0..b30fe39ab8
Review mode: full
Session key: repo-1308402561-pr-6-v2
Generation key: 5e27bc436108b890287c303dc8622a61c8892f4b49322109c50ffb9f07ffeb0f
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 24s
Input tokens: 168,412
Cached input tokens: 124,115
Cache write tokens: 44,285
Cache hit ratio: 73.7%
Output tokens: 1,687
Reasoning tokens: 1,110
Total tokens: 170,099
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total |
|---|---|---|---|---|---|---|---|---|
chunk:1/1:b60e7da89aaabff7a165a5fc758b1b967cf58ae77d7d85923061d0c27a99aa61 |
17s | 80,385 | 37,774 | 47.0% | 42,605 | 1,244 | 904 | 81,629 |
aggregate:5e27bc436108b890287c303dc8622a61c8892f4b49322109c50ffb9f07ffeb0f |
7s | 88,027 | 86,341 | 98.1% | 1,680 | 443 | 206 | 88,470 |
One behavior regression found.
1 inline finding published.
There was a problem hiding this comment.
🤖 OpenAI PR review
Reviewed commit: b30fe39ab8
Reviewed range: a622026ca0..b30fe39ab8
Review mode: reused
Session key: repo-1308402561-pr-6-v2
Generation key: 5e27bc436108b890287c303dc8622a61c8892f4b49322109c50ffb9f07ffeb0f
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 24s
Input tokens: 168,412
Cached input tokens: 124,115
Cache write tokens: 44,285
Cache hit ratio: 73.7%
Output tokens: 1,687
Reasoning tokens: 1,110
Total tokens: 170,099
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total |
|---|---|---|---|---|---|---|---|---|
chunk:1/1:b60e7da89aaabff7a165a5fc758b1b967cf58ae77d7d85923061d0c27a99aa61 |
17s | 80,385 | 37,774 | 47.0% | 42,605 | 1,244 | 904 | 81,629 |
aggregate:5e27bc436108b890287c303dc8622a61c8892f4b49322109c50ffb9f07ffeb0f |
7s | 88,027 | 86,341 | 98.1% | 1,680 | 443 | 206 | 88,470 |
One behavior regression found.
1 inline finding published.
There was a problem hiding this comment.
🤖 OpenAI PR review
Reviewed commit: 585315c766
Reviewed range: b30fe39ab8..585315c766
Review mode: incremental
Session key: repo-1308402561-pr-6-v2
Generation key: 4a40458e8682a5bc4e13bc3c25b77b20943cb1b2b0291b9031ccc2f7cf53db8f
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 7s
Input tokens: 184,761
Cached input tokens: 182,079
Cache write tokens: 2,670
Cache hit ratio: 98.5%
Output tokens: 446
Reasoning tokens: 37
Total tokens: 185,207
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total |
|---|---|---|---|---|---|---|---|---|
chunk:1/1:f3fc9982671e7d08e3a06571b07e52f669e5b78a8fc5e429cd6a10b0a72a0caa |
4s | 91,262 | 89,233 | 97.8% | 2,023 | 289 | 30 | 91,551 |
aggregate:4a40458e8682a5bc4e13bc3c25b77b20943cb1b2b0291b9031ccc2f7cf53db8f |
3s | 93,499 | 92,846 | 99.3% | 647 | 157 | 7 | 93,656 |
No actionable findings in this generation.
No actionable findings.
1 participant
Summary
v0.2tag and the Chatroom, Pet, PetDef, provider, and RuntimeProfile example boundaries unchangedValidation
go test ./pkgs/gizclaw/api/apitypes -count=1in a clean GizClaw worktree at merged #528workflows/throughapitypes.Resourcegit diff --checkThe immutable
v0.2tag remains untouched. Publishv0.2.1from the merged commit after review.Closes #5